Section 1.3: A Simple Inventory System

Transcription

1 Section 1.3: A Simple Inventory System Discrete-Event Simulation: A First Course c 2006 Pearson Ed., Inc Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 1/ 27

2 Section 1.3: A Simple Inventory System customers. demand items.. facility. order items.. supplier Distributes items from current inventory to customers Customer demand is discrete Simple one type of item Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 2/ 27

3 Inventory Policy Transaction Reporting Inventory review after each transaction Significant labor may be required Less likely to experience shortage Periodic Inventory Review Inventory review is periodic Items are ordered, if necessary, only at review times (s, S) are the min,max inventory levels, 0 s < S We assume periodic inventory review Search for (s,s) that minimize cost Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 3/ 27

4 Conceptual Model Inventory System Costs Holding cost: for items in inventory Shortage cost: for unmet demand Setup cost: fixed cost when order is placed Item cost: per-item order cost Ordering cost: sum of setup and item costs Additional Assumptions Back ordering is possible No delivery lag Initial inventory level is S Terminal inventory level is S Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 4/ 27

5 Specification Model Time begins at t = 0 Review times are t = 0,1,2,... l i 1 : inventory level at beginning of i th interval o i 1 : amount ordered at time t = i 1, (o i 1 0) d i : demand quantity during i th interval, (d i 0) Inventory at end of interval can be negative Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 5/ 27

6 Inventory Level Considerations Inventory level is reviewed at t = i 1 If at least s, no order is placed If less than s, inventory is replenished to S { 0 li 1 s o i 1 = S l i 1 l i 1 < s Items are delivered immediately At end of i th interval, inventory diminished by d i l i = l i 1 + o i 1 d i Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 6/ 27

7 Time Evolution of Inventory Level Algorithm l 0 = S; /* the initial inventory level is S */ i = 0; while (more demand to process ) { i++; if (l i 1 < s) o i 1 = S - l i 1 ; else o i 1 = 0; d i = GetDemand(); l i = l i 1 + o i 1 - d i ; } n = i; o n = S - l n ; l n = S; /* the terminal inventory level is S */ return l 1, l 2,..., l n and o 1, o 2,..., o n ; Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 7/ 27

8 Example 1.3.1: SIS with Sample Demands Let (s, S) = (20, 60) and consider n = 12 time intervals i : d i : l i 80 S 40 s i Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 8/ 27

9 Output Statistics What statistics to compute? Average demand and average order d = 1 n n i=1 d i ō = 1 n n o i. i=1 For Example data d = ō = 305/ items per time interval Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 9/ 27

10 Flow Balance Average demand and order must be equal Ending inventory level is S Over the simulated period, all demand is satisfied Average flow of items in equals average flow of items out customers. d facility. ō supplier The inventory system is flow balanced Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 10/ 27

11 Constant Demand Rate Holding and shortage costs are proportional to time-averaged inventory levels Must know inventory level for all t Assume the demand rate is constant between review times l(t) 80 S 40 s t Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 11/ 27

12 Inventory Level as a Function of Time The inventory level at any time t in i th interval is l(t) = l i 1 (t i + 1)d i if demand rate is constant between review times l i 1 (i 1, l i 1 ) l(t) l i 1 d i l(t) = l i 1 (t i + 1)d i (i, l i 1 d i). i 1 l i 1 = l i 1 + o i 1 represents inventory level after review i t Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 12/ 27

13 Inventory Decrease Is Not Linear Inventory level at any time t is an integer l(t) should be rounded to an integer value l(t) is a stair-step, rather than linear, function l(t) l i 1 l i 1 d i (i 1, l i 1 ) l(t) = l i 1 (t i + 1)d i (i, l i 1 d i). i 1 i t Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 13/ 27

14 Time-Averaged Inventory Level l(t) is the basis for computing the time-averaged inventory level Case 1: If l(t) remains non-negative over i th interval i l+ i = l(t)dt i 1 Case 2: If l(t) becomes negative at some time τ τ l+ i = l(t)dt i 1 l i i = l(t)dt τ l + i l i is the time-averaged holding level is the time-averaged shortage level Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 14/ 27

15 Case 1: No Back Ordering No shortage during i th time interval iff. d i l i 1 i 1 i 0 l i 1 d i l i 1 l(t) t. (i 1, l i 1 ) (i, l i 1 d i) Time-averaged holding level: area of a trapezoid l+ i = i i 1 l(t)dt = l i 1 + (l i 1 d i) 2 = l i d i Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 15/ 27

16 Case 2: Back Ordering Inventory becomes negative iff. d i > l i 1 i 1 τ i l i 1 d i 0 s l i 1 l(t) t. (i 1, l i 1 ) (i, l i 1 d i) Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 16/ 27

17 Case 2: Back Ordering (Cont.) l(t) becomes negative at time t = τ = i 1 + (l i 1 /d i) Time-averaged holding and shortage levels for i th interval computed as the areas of triangles l i τ l+ i = i 1 i = τ l(t)dt = = (l i 1 )2 2d i l(t)dt = = (d i l i 1 )2 2d i Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 17/ 27

18 Time-Averaged Inventory Level Time-averaged holding level and time-averaged shortage level l+ = 1 n n i=1 l+ i l = 1 n n i=1 l i Note that time-averaged shortage level is positive The time-averaged inventory level is l = 1 n n 0 l(t)dt = l + l Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 18/ 27

19 Computational Model sis1 is a trace-driven computational model of the SIS Computes the statistics d, ō, l +, l and the order frequency ū ū = number of orders n Consistency check: compute ō and d separately, then compare Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 19/ 27

20 Example 1.3.4: Executing sis1 Trace file sis1.dat contains data for n = 100 time intervals With (s,s) = (20,80) ō = d = ū = 0.39 l+ = l = 0.25 After Chapter 2, we will generate data randomly (no trace file) Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 20/ 27

21 Operating Costs A facility s cost of operation is determined by: c item : unit cost of new item c setup : fixed cost for placing an order c hold : cost to hold one item for one time interval c short : cost of being short one item for one time interval Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 21/ 27

22 Case Study Automobile dealership that uses weekly periodic inventory review The facility is the showroom and surrounding areas The items are new cars The supplier is the car manufacturer...customers are people convinced by clever advertising that their lives will be improved significantly if they purchase a new car from this dealer. (S. Park) Simple (one type of car) inventory system Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 22/ 27

23 Example 1.3.5: Case Study Materialized Limited to a maximum of S = 80 cars Inventory reviewed every Monday If inventory falls below s = 20, order cars sufficient to restore to S For now, ignore delivery lag Costs: Item cost is c item = $8000 per item Setup cost is c setup = $1000 Holding cost is c hold = $25 per week Shortage cost is c hold = $700 per week Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 23/ 27

24 Per-Interval Average Operating Costs The average operating costs per time interval are item cost : c item ō setup cost : c setup ū holding cost : c hold l + shortage cost : c short l The average total operating cost per time interval is their sum For the stats and costs of the hypothetical dealership: item cost : $ = $234, 320 per week setup cost : $ = $390 per week holding cost : $ = $1, 060 per week shortage cost : $ = $175 per week Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 24/ 27

25 Cost Minimization By varying s (and possibly S), an optimal policy can be determined Optimal minimum average cost Note that ō = d, and d depends only on the demands Hence, item cost is independent of (s,s) Average dependent cost is avg setup cost + avg holding cost + avg shortage cost Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 25/ 27

26 Experimentation Let S be fixed, and let the demand sequence be fixed If s is systematically increased, we expect: average setup cost and holding cost will increase as s increases average shortage cost will decrease as s increases average dependent cost will have U shape, yielding an optimum From results (next slide), minimum cost is $1550 at s = 22 Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 26/ 27

27 Example 1.3.7: Simulation Results setup cost s holding cost s shortage cost total cost s s Discrete-Event Simulation: A First Course Section 1.3: A Simple Inventory System 27/ 27